Zoho founder Sridhar Venbu shared concerns about the growing use of AI-generated code in software development, saying the technology could increase the amount of code developers generate without commensurate gains in actual productivity. In a post on X (formerly Twitter), Vembu reposted a comment by François Chollet, founder of the Intelligence Sciences Institute NDEA. He argued that even though developers are shipping far more code with the help of AI tools, the actual value created hasn’t increased that much. The discussion reflects a broader conversation in the tech industry as companies invest billions of dollars in AI coding assistants and developer tools.
AI has little contribution to developer productivity
Chollet suggested in his post that the actual gains may be much smaller. “The amount of code developers ship has increased approximately 10 times,” Chollet wrote in X. “However, developers’ net productivity (value produced per unit of time) has increased very little, if at all.” He added that one reason for this is that developers may now be solving smaller or more incremental problems using the additional code that is generated. Chollet also said that some of the newly generated code can itself cause additional problems that developers will need to fix later.“The bigger part is that the new code is creating problems of its own,” he wrote.Reacting to the post, Vembu highlighted a comment on Hacker News that said code generated with large-scale language models is “applicable.” [an] Despite having an incredible amount of knowledge, they have virtually no understanding of the problem at all. ”According to Venbu, this comment captures what he called the “developer productivity paradox” currently facing the software industry.
what is this
“It can be applied to [an] Despite having an incredible amount of knowledge, they understand virtually nothing about the problem.”Comment from Hacker News regarding large-scale code generated with LLM.This and the post cited below encapsulate the developer productivity paradox that the software industry is grappling with today.Hundreds of billions of dollars are being invested on the premise that software engineering productivity will increase by more than 10x. At least so far, productivity gains have been modest.There will be interesting times ahead.The discussion comes as major technology companies are investing heavily in AI systems designed to help developers write code faster. Many companies believe that AI tools can improve software engineering productivity by as much as 10x.
